Welcome     Marketing & Loyalty     Speaking & Connecting     Covid & Other     Books by Brian     Contact Brian  

What is loyalty?

By Brian Woolf
July 30, 2002

A critique of the Harvard Business Review article, 'The Mismanagement of Customer Loyalty'

Introduction A lot of welcome evaluation of loyalty marketing has been printed in recent months. This includes an article by Professors Werner Reinartz and V. Kumar published in the Harvard Business Review (July 2002). Because of the credence associated with this magazine it deserves a careful review.

The key message of the article, The Mismanagement of Customer Loyalty, is set out in the final paragraph: ""No company should ever take for granted the idea that managing customers for loyalty is the same as managing them for profits." In other words, they are stating that focusing solely on customer loyalty does not always lead to increased profits.

That is, of course, a truism. Whether our profits increase or decrease with increased loyalty depends upon how much we spend on building our customers' loyalty versus the resultant incremental gain from those expenditures. Frankly, I am unaware of any loyalty advocate who suggests that every expenditure aimed at building customer loyalty will have a positive ROI.

The statement is akin to the idea that if you increase sales you will increase profits. If this were true, then every company would lower prices by 10%, thereby increasing sales. But this sales increase won't necessarily result in higher profits. As in the loyalty statement, there is an implied qualifier: increased sales will lead to increased profits, provided that the incremental gains are achieved in a cost-effective fashion.

We would all hope that every company does a cost/benefit projection before committing to any significant customer loyalty- or sales-building program. The research behind the article The major part of the article revolves around the authors' research showing that ""the relationship between loyalty and profitability is much weaker - and subtler - than the proponents of loyalty programs claim." They state that they have ""discovered little or no evidence to suggest that customers who purchase steadily from a company over time [their definition of loyalty] are necessarily cheaper to serve, less price sensitive, or particularly effective at bringing in new business." Their findings come from a study of 16,000 customers in four companies - all, presumably, with a formal loyalty program (but no details are provided). The companies were:
  ·  A French retail food company
  ·  A German direct brokerage house
  ·  A large US mail-order company
  ·  A US high-tech corporate service provider Another viewpoint The authors claim that their findings were consistent in all four companies listed. As the majority of my work over the past ten years has been with loyalty card food retailers on five continents and, as one of the four companies in the authors' study was a French food retailer, it is fair to compare my observations with those of the authors. Loyalty drivers Let us begin by understanding that the key loyalty drivers of a business are what McDonald's call QSVC: Quality, Service, Value, and Cleanliness. For non-retailers, the 'C' would more aptly stand for Convenience (eg, how easy is it to place a catalog order or a stock trade). Until those four elements are in good shape, it's probably best to defer the introduction of a loyalty program.

In my experience, the primary role of a retailer loyalty card is to gather data about customers. This in turn leads to customer comprehension and cost insights (eg, customer retention rates at different spending levels, response rates to offers, new customer conversion rates, and where money is being wasted on circulars), followed by appropriate marketing action and follow-up analysis. In other words, a loyalty card is primarily an information card from which follows better decision-making in both marketing and cost-reduction.

I will never forget the conversations I had with the management of two major retailers (one in Europe and one in the US). They told me, on separate occasions, that the major beneficiary of their loyalty card programs was not their marketing departments but their real estate departments! The information generated by their customer cards showed them, for example, not just how far customers travelled to their stores but also how much customers' spending changed as the distance varied. This was of significant benefit to them as they evaluated new store sites.

The best way, therefore, to view a loyalty card is to ask yourself: How can the resultant customer information be used to help the company achieve its strategic goals? (That includes loyalty and profit goals, of course.) This is certainly the hallmark of the best companies with loyalty programs I have seen around the world - some of which can be read about in my book, Loyalty Marketing: The Second Act. What is Loyalty? There is no universally accepted definition of loyalty and, therefore, who loyal customers are. Some claim we should measure customer loyalty by their share of wallet; others say it should be based on the customer retention rate; others push frequency as the best measure, while others claim it's the customers' attitude towards the company that best describes loyalty.

Understanding which definition is used in The Mismanagement of Customer Loyalty is crucial because if a reader's understanding of a loyalty program is materially different from that used by Reinartz and Kumar, the article's arguments and conclusions may have no relevance to how the reader practices loyalty marketing. This is certainly true in my case. I don't accept their ""customers who purchase steadily from a company over time" (with no other qualifiers) as a good definition. In fact, I don't know any expert or practitioner who does.

One of the challenges practitioners have had is to find a definition that is easy to understand and easy to measure. Otherwise it's just a theoretical abstract. Over the past decade my retail clients have evolved, now considering their loyalty programs as sources of data to help accomplish their corporate goals.

To achieve this, identifiable customers are sorted into one of five classes: Diamonds, Rubies, Opals, Pearls or New. This method of classification is described using the acronym DROP'N. Among food retailers, the quarterly breakpoints most often used are 1300, 650, and 325 US Dollars (or Euros, depending on whether it's a European or American retailer). In other words, a Diamond's average weekly spend in a quarter is over 100, a Ruby's is from 50 to 100, an Opal's is from 25 to 50, and a Pearl's is under 25 (all in either US$ or Euros). New customers are those who, during the quarter, have bought for the first time. The goal of these retailers is to increase the number of its 'Best Customers', comprising the two top categories (Diamonds and Rubies). The reasoning is simple: As a general rule, retailers have found that Best Customers have:
  ·  The lowest defection rates
  ·  The lowest processing costs
  ·  The highest gross profit yields

The authors state that they did not find these correlations in the French food retailer (and the three non-retailers). That, I suggest, is because of their definition of loyalty. Spending levels in a shopping cycle - quarterly, for example - is a more practical measure. Is a farmer who shops infrequently but spends heavily on each visit less loyal that someone who shops frequently but spends a small amount each visit? Surely, it's the end result - spending - that counts from a business viewpoint.

The authors defined loyalty as 'frequency'. It follows that their best customers must be those who buy most frequently. Unfortunately this includes a number of low-spending customers whose spending characteristics and economics can be quite different from the frequent, high-spending customers (the Diamonds and Rubies). Best Customers - based on spending Let me share what my experiences show when loyalty is defined as the amount spent, regardless of the buyer's frequency. These conclusions are made after working with, and being privy to, company results comprising many millions of customers.

A good loyalty program at a food retailer captures over 80% of sales on its card, and the average number of active households per store ranges from 10,000 to 25,000 (where the store size may range from 30,000 to 80,000 square feet). Thus, it's common to see a chain of just 50 stores having at least 500,000 (or as many as a million) customers. Some chains (such as Tesco and Boots in the UK) have over 8 million active members.

In contrast, the total research base used in the article comprised 16,000 customers for the four companies. Let's assume that the French food retailer contributed half of that total (8,000 customers). This raises questions regarding the extensiveness of the research. Did they just study a small, single-store food retailer, or did they study a chain that was only capturing a small percentage of sales on its card? If the latter, then it's fair to wonder how representative the research data really is. Best Customers have the lowest defection rates Set out below we see the defection rate of a large US-based food retailer several years after its loyalty program had been established. The company had over one million cardholders, with over 80% of sales being captured on the card.

The table shows the customer composition of each segment in a quarter, and the defections from each segment in the same quarter of the following year. It is representative of many I have studied. The base is expressed here in terms of 10,000 customers to make easy visual calculations and to disguise the company's identity.

  Average spend
per week ($)
Base year
(# households)
(# households)
rate (%)
Diamonds > 100 302 8 2.6
Rubies 50 - 100 1,175 53 4.5
Best customers: > 50 1,477 61 4.1
Opals 25 - 50 1,635 127 7.8
Pearls < 25 4,928 835 16.9
Pre-existing actives:   8,040 1,023 12.7
New customers   1,960 550 28.1
TOTAL:   10,000 1,573 15.7
Table 1: Customer defection rates

As Table 1 shows, the higher the average customer spending per week, the lower the (annual) defection rate. In this company's case, only 4.1% of its Best Customers had defected a year later, while 16.9% of its lowest spending (Pearl) customers had. It follows that Best Customers (where 'best' is defined by spending, not frequency) have the highest lifetime value of sales (as they have both the highest spending and retention rates). Best Customers have the lowest processing costs Let me give two examples to explain why this is so among food retailers:

1) Lower per-item checkout costs
The following table shows the actual results for another major US-based food retailer. It shows a quarter's results for all active households, broken into quintiles. Quintile #1 is the top 20% of spenders, while quintile #5 represents the bottom 20%. The table assumes that the average item sold was US$2.00.

Quintile Average spend
in quarter ($)
Average spend
per visit ($)
Average items
per visit (#)
#1 891 46.99 23.5
#2 359 33.10 16.6
#3 175 26.70 13.4
#4 79 21.42 10.7
#5 24 12.54 6.3
All households: 307 36.44 18.2
Table 2: Customer spending profile

This table shows that, the more a customer spends on average each week in a quarter, the more he or she spends, on average, per visit. Top quintile customers spent US$46.99 on each visit compared with only US$12.54 by the bottom quintile customers. It follows that the average number of items in each transaction is correspondingly higher for top quintile customers and lower in the lower quintiles. The table uses a typical average selling price of US$2.00 per item for all customers to illustrate this point.

Now, there are two types of transactions at the checkout: the process time per item (ringing and bagging) and the fixed time (greeting and thanking the customer, receiving the payment, and giving change for the transaction). The former type is basically the same time per item processed, whilst the latter is largely consistent regardless of the number of items in the entire transaction. Dividing the fixed time over 23.5 items in a transaction (as in the top quintile) means a lot lower cost per item than when divided over a bottom quintile transaction (only 6.3 items). For a large retailer, this can represent a significant difference in costs.

2) Lower bad check losses
In the US, a large number of retail payments are still made by check (British: cheque). One outstanding retailer, after its loyalty program had been in place for four years, decided to see if there was a correlation between bad checks written off and the spending level of its customers. The following table is a summary of what was discovered:

Quintile Total of
bad checks
Total of
company's sales
Bad checks/sales
#1 4.7% 56.1% 8%
#2 7.4% 19.6% 38%
#3 10.3% 9.1% 113%
#4 11.1% 4.1% 271%
#5 28.0% 1.1% 2545%
Cardholders: 61.5% 90.0% 68%
NOIDs 38.5% 10.0% 385%
TOTAL: 100% 100% 100%
Table 3: Bad check losses (by quintile)

Again, we see how economics favour our Best Customers (as defined by spending) in retailing. The company captured 90% of its total sales on its loyalty card. The other 10% came from customers with no identification (which are called NOIDs, for short). The top 20% of cardholders accounted for only 4.7% of the bad checks written off, yet they generated 56.1% of the company's total sales.

As we drop to each lower decile, the share of bad checks increases while the share of sales declines. The column showing the relationship between bad checks and sales clearly illustrates how higher spending customers created a much smaller percentage of bad checks than low and unidentified spenders. (As a result of these findings, the company made significant changes to its check cashing policy, easing up on better customers and imposing tighter restrictions on lower spending and unidentified customers.)

Parenthetically, just two months ago, a senior Tesco executive stated (at a forum in London) that the company was finding that Best Customers who held Tesco insurance policies were incurring claims at a lower rate than other customers. To me, it was yet another example of the favourable economics of Best Customers. Best Customers yield a higher Gross Profit % The authors said they found no correlation between Best Customers and higher prices. Do they mean higher prices, or the end result of higher prices (a higher gross profit percentage)? In almost every company where I have sorted customers according to spending levels (after a loyalty program is installed) I have found a positive correlation between spending levels and the gross profit percentage earned on those sales.

One such example (a food retailer) is shown below. This table shows the total markdown percentage on sales to Diamonds, Rubies, Opals and Pearls. (The markdown is the amount of price reductions given.) Every item that was reduced in price, either on a short or long term basis, was captured. Obviously, the bigger the markdown percentage, the lower the gross profit percentage. Like most food retailers who do not have reliable item cost data, this company captures its gross profit weekly using this markdown method.

Results for
the Quarter
as % of sales
Diamonds 10.1%
Rubies 11.9%
Opals 13.3%
Pearls 15.0%
TOTAL: 12.4%
Table 4: Price markdowns (by quintile)

Clearly, the Best Customers were generating a higher gross profit percentage on sales than the lower spending customers. No, not because of raising prices to higher spenders but because Best Customers have a different sales mix. The company's larger-order shopping baskets simply reflect that these customers are buying there for most of their needs. This includes the higher-margin, slower-moving items. This also means that the share of lower margin 'specials' in their baskets is less. Combined, the result is a higher gross profit percentage.

However, once the loyalty leaders discover this correlation, I find that they typically go about lowering the gross margin of the Best Customers (through additional special prices and privileges, for example) while raising it on the Opals and Pearls (eg, with fewer special offers and the imposition of quantity limits). They have found that this approach improves both the spending and retention of their Best Customers (who are, of course, the foundation of their prosperity). The goal of such leading retailers is to keep increasing, in a cost-effective manner, the number of Best Customers per store, as that is the surest way to build both short-term and long-term success.

There is only one exception I have found to the above situation: that of an aggressive price discounter. In this case, a significant number of its Best Customers are so price-driven (which is the primary attraction to the company) that their gross profit percentage is not necessarily higher than the lower spending customer classes. Word-of-mouth The authors also challenged the concept of word-of-mouth as a by-product of loyalty. Obviously, customers at all spending levels will have varying degrees of preference for a company. To me, the most powerful indicator of the power of word-of-mouth has been in two food retailers that dropped their weekly newspaper advertisements more than five years ago, and have continued to grow and succeed ever since. Where did Dorothy Lane Market in Dayton, Ohio, and Superquinn in Dublin, Ireland, get their new customers? One has to assume that world-of-mouth played a key role. (Their stories can be read in Loyalty Marketing: The Second Act.) RFM The final major point made in The Mismanagement of Customer Loyalty relates to RFM (Recency, Frequency and Monetary Value). The authors state, after showing various research, that ""RFM is a poor way to measure loyalty"". Without question! But again, I am unaware of any practitioners who use it that way.

RFM is an excellent tactical targeting tool with a proven record of some fifty years - but it is not a loyalty measurement tool because it's a moving target. Successful retailers have found the fixed threshold DROP'N customer classification approach to be the most practical method at present. The bottom line Given the definition of loyalty that the authors use - the frequency of customers (with no qualification) - their case may be valid. However, it is a definition I have not seen in practice during my ten years of work with retailers around the world. The most successful definition of loyalty that I have seen among food retailers uses spending as the criteria.

Despite the authors' claims that they couldn't find the benefits of loyalty marketing, those benefits certainly do exist as the examples above clearly show. It's all in your definition of loyalty, for that tells you where to look.

The moral, to me, is that when someone makes assertions about loyalty - either for or against - you must carefully learn what he or she means when using that simple word. It just might not have the same meaning that you give it ... and, therefore, the assertions may not apply to your situation. Perhaps the mismanagement of loyalty sometimes occurs because of the misdefinition of loyalty.

The original article, The Mismanagement of Customer Loyalty, can be purchased online from the Harvard Business Review, at

Copyright © 2002 - 2022 Brian Woolf

About the author...

Brian Woolf is a global leader in loyalty marketing and has written three definitive works on the subject, Measured Marketing: A Tool to Shape Food Store Strategy, Customer Specific Marketing, and Loyalty Marketing: The Second Act. He devotes his time to helping retailers develop, critique and strengthen their loyalty programs.

The techniques and metrics Brian Woolf has developed have become guiding principles for those operating some of the world's most successful programs. He is the President of the Retail Strategy Center, and has consulted, and spoken at conferences, in the US, Europe, Japan, and Australasia.

Prior to his total commitment to loyalty marketing, his corporate roles included Deputy Managing Director of Progressive Enterprises, a major New Zealand retailer; and Chief Financial Officer of Food Lion, a leading US food retailer. He has an M.Com. (Economics) from the University of Auckland, New Zealand, and an MBA from the Harvard Business School.

Copyright © 2022 Brian Woolf